Until the Middle of November S&P-500 Might Be Still Good

The stock market during a month ahead might be driven by a few expected fundamental news. Nevertheless, technical factors can give some additional momentum for the market. In October, S&P-500 gained more than 3%. The last four weeks, S&P-500 weekly performance was 1.65%, 0.95%, 0.59%, and 0.02%. The trend does not look optimistic. However, according to a cycle analysis, October’s uptrend cycle can be still intact. The chart shows a possible resumption of uptrend until November 11:

The forecast was calculated using software tool SMAP-3. A cycle prediction is based on the hypothesis that statistically revealed cycles may be repeatable in the future and, therefore, they can be used to build the summarized curve of future movements.

Stock Price Forecast by Pattern Similarity

Pattern recognition systems are able to classify data based on statistical information extracted from the patterns. The patterns should be classified in order to extract useful information. Normally, a pattern recognition system consists of the input, data converter, an extraction processor to classify accordingly to descriptive scheme, and output grouped data. The most popular systems are based on statistical pattern recognition, i.e., statistical classification of patterns, assuming that the patterns are generated by a probabilistic system.

If we assume that some patterns are repeatable in the future, we can use selected present patterns to predict the future ones. Investment Analyzer (IA) by Addaptron Software has an extra feature to predict prices of the selected stock (index) using pattern similarity. The prediction period can be chosen from 1 day to around 60 trading days; the period of historical data that used for matching should be in 4..16 times longer than prediction period. IA searches for the best match from the internal database by scanning all historical data.

IA ranks all possible matches on the basis of minimum deviation and maximum correlation within given historical period. IA performs pattern matching using open, high, low, and close prices and volume data. When scanning is completed, IA composes forecast using several best matched patterns (top ranked). Since the statistical regularities of the patterns help to create more stable picture, IA allows adding up many top-rated patterns. The composite result is built as a weighted average with weights proportionally patterns’ ranks. The number of patterns that form composite forecast can be adjusted; the number range is 1..99.

The chart below has been plotted using IA pattern similarity feature. SP-500 index prices have been used as input; the output is predicted prices for two weeks of April 2010:

© Alex Shmatov. Published with permission of the copyright owner. Further reproduction strictly prohibited without permission.

Using Logarithmic Price Scale for Stock Performance

It is nice to see a simple linear chart with dollar numbers that is easy to calculate. However, linear price scale may not be good for significant changes in stock prices if we calculate money in terms of return on investment (ROI). For example, the change $10 of SP-500 gives around 10% market performance in 1970 but the same $10 change is equivalent only 1% of SP-500 performance in 2010.

So that be aware that big changes on linear graph can be misleading since they may not represent big ROI. That is why logarithmic (log) price scales are used by many investors and technical traders. Log scale graphs correctly show in the percentage the rate of return-on-investment for both, small and big changes in prices. Log graphs show percentage changes accurately, since the same interval anywhere on the price scale represents the same percentage change.

Such software tool, as Stock Market Analyzer-Predictor SMAP-2/3, works with multi-year periods. That is why it uses a logarithmic re-normalization.

© Alex Shmatov. Published with permission of the copyright owner. Further reproduction strictly prohibited without permission.